Abstract-Architecture Compliance Checking (ACC) is an approach to verify the conformance of implemented program code to high-level models of architectural design. ACC is used to prevent architectural erosion during the development and evolution of a software system. Static ACC, based on static software analysis techniques, focuses on the modular architecture and especially on rules constraining the modular elements. A semantically rich modular architecture (SRMA) is expressive and may contain modules with different semantics, like layers and subsystems, constrained by rules of different types. To check the conformance to an SRMA, ACC-tools should support the module and rule types used by the architect. This paper presents requirements regarding SRMA support and an inventory of common module and rule types, on which basis eight commercial and non-commercial tools were tested. The test results show large differences between the tools, but all could improve their support of SRMA, what might contribute to the adoption of ACC in practice.
Abstract-Architecture Compliance Checking (ACC) is useful to bridge the gap between architecture and implementation. ACC is an approach to verify conformance of implemented program code to high-level models of architectural design. Static ACC focuses on the modular software architecture and on the existence of rule violating dependencies between modules. Accurate tool support is essential for effective and efficient ACC. This paper presents a study on the accuracy of ACC tools regarding dependency analysis and violation reporting. Seven tools were tested and compared by means of a custom-made test application. In addition, the code of open source system Freemind was used to compare the tools on the number and precision of reported violation and dependency messages. On the average, 74 percent of 34 dependency types in our custom-made test software were reported, while 69 percent of 109 violating dependencies within a module of Freemind were reported. The test results show large differences between the tools, but all tools could improve the accuracy of the reported dependencies and violations.
